Search

Real-time visual feedback from catheterization analysis is crucial for enhancing surgical safety and efficiency during endovascular interventions. However, existing datasets are often limited to specific tasks, small scale, and lack the comprehensive annotations necessary for broader endovascular intervention understanding. To tackle these limitations, we introduce CathAction, a large-scale dataset for catheterization understanding. Our CathAction dataset encompasses approximately 500,000 annotated frames for catheterization action understanding and collision detection, and 25,000 ground truth masks for catheter and guidewire segmentation. For each task, we benchmark recent related works in the field. We further discuss the challenges of endovascular intentions compared to traditional computer vision tasks and point out open research questions. We hope that CathAction will facilitate the development of endovascular intervention understanding methods that can be applied to real-world applications. The dataset is available at https://airvlab.github.io/cathaction/., Comment: 10 pages. Webpage: https://airvlab.github.io/cathaction/

Brain tumor segmentation is often based on multiple magnetic resonance imaging (MRI). However, in clinical practice, certain modalities of MRI may be missing, which presents a more difficult scenario. To cope with this challenge, Knowledge Distillation, Domain Adaption, and Shared Latent Space have emerged as commonly promising strategies. However, recent efforts typically overlook the modality gaps and thus fail to learn important invariant feature representations across different modalities. Such drawback consequently leads to limited performance for missing modality models. To ameliorate these problems, pre-trained models are used in natural visual segmentation tasks to minimize the gaps. However, promising pre-trained models are often unavailable in medical image segmentation tasks. Along this line, in this paper, we propose a novel paradigm that aligns latent features of involved modalities to a well-defined distribution anchor as the substitution of the pre-trained model}. As a major contribution, we prove that our novel training paradigm ensures a tight evidence lower bound, thus theoretically certifying its effectiveness. Extensive experiments on different backbones validate that the proposed paradigm can enable invariant feature representations and produce models with narrowed modality gaps. Models with our alignment paradigm show their superior performance on both BraTS2018 and BraTS2020 datasets.

This paper reports recent lattice characterization results obtained at the National Synchrotron Light Source II (NSLS-II) storage ring, conducted without reliance on a lattice model. A pair of beam position monitors (BPMs) with bunch-by-bunch (B$\times$B) resolution, were recently installed in a section of the storage ring free of magnetic fields. The new BPM pair measured the beam, or bunch's transverse Poincar\'e map precisely after the beam was excited. Linear one-turn-matrices (OTM) were then derived, and from these, the 4-dimensional coupled Twiss parameters were extracted at the locations of the BPM pair. By normalizing beam oscillation amplitudes with the Twiss parameters, the global action-variables were obtained. These action-variables facilitated the measurement of the local Twiss parameters observed by other BPMs independent on lattice model. This method is general, and particularly useful in certain scenarios such as a round beam mode in a diffraction-limited light source ring. We applied it to assess both weakly and strongly coupled lattices at the NSLS-II ring. Through analysis of the strongly coupled lattice, the quadrupole tilt errors were estimated to be less than 400 \si{\mu}rad. Utilizing the BPMs' B$\times$B resolution, for the first time we observed the variations of the linear lattice along a long bunch-train., Comment: 28 pages, 16 figures, accepted by Nuclear Inst. and Methods in Physics Research, A

In TDD mmWave massive MIMO systems, the downlink CSI can be attained through uplink channel estimation thanks to the uplink-downlink channel reciprocity. However, the channel aging issue is significant under high-mobility scenarios and thus necessitates frequent uplink channel estimation. In addition, large amounts of antennas and subcarriers lead to high-dimensional CSI matrices, aggravating the pilot training overhead. To systematically reduce the pilot overhead, a spatial, frequency, and temporal domain (3D) channel extrapolation framework is proposed in this paper. Considering the marginal effects of pilots in the spatial and frequency domains and the effectiveness of traditional knowledge-driven channel estimation methods, we first propose a knowledge-and-data driven spatial-frequency channel extrapolation network (KDD-SFCEN) for uplink channel estimation by exploiting the least square estimator for coarse channel estimation and joint spatial-frequency channel extrapolation to reduce the spatial-frequency domain pilot overhead. Then, resorting to the uplink-downlink channel reciprocity and temporal domain dependencies of downlink channels, a temporal uplink-downlink channel extrapolation network (TUDCEN) is proposed for slot-level channel extrapolation, aiming to enlarge the pilot signal period and thus reduce the temporal domain pilot overhead under high-mobility scenarios. Specifically, we propose the spatial-frequency sampling embedding module to reduce the representation dimension and consequent computational complexity, and we propose to exploit the autoregressive generative Transformer for generating downlink channels autoregressively. Numerical results demonstrate the superiority of the proposed framework in significantly reducing the pilot training overhead by more than 16 times and improving the system's spectral efficiency under high-mobility scenarios., Comment: 13 pages, 11 figures, 3 tables. This paper has been submitted to IEEE journal for possible publication

Large Language Models (LLMs) have significantly advanced the field of Natural Language Processing (NLP), achieving remarkable performance across diverse tasks and enabling widespread real-world applications. However, LLMs are prone to hallucination, generating content that either conflicts with established knowledge or is unfaithful to the original sources. Existing hallucination benchmarks primarily focus on sentence- or passage-level hallucination detection, neglecting dialogue-level evaluation, hallucination localization, and rationale provision. They also predominantly target factuality hallucinations while underestimating faithfulness hallucinations, often relying on labor-intensive or non-specialized evaluators. To address these limitations, we propose HalluDial, the first comprehensive large-scale benchmark for automatic dialogue-level hallucination evaluation. HalluDial encompasses both spontaneous and induced hallucination scenarios, covering factuality and faithfulness hallucinations. The benchmark includes 4,094 dialogues with a total of 146,856 samples. Leveraging HalluDial, we conduct a comprehensive meta-evaluation of LLMs' hallucination evaluation capabilities in information-seeking dialogues and introduce a specialized judge language model, HalluJudge. The high data quality of HalluDial enables HalluJudge to achieve superior or competitive performance in hallucination evaluation, facilitating the automatic assessment of dialogue-level hallucinations in LLMs and providing valuable insights into this phenomenon. The dataset and the code are available at https://github.com/FlagOpen/HalluDial.

The evaluation of Long Video Understanding (LVU) performance poses an important but challenging research problem. Despite previous efforts, the existing video understanding benchmarks are severely constrained by several issues, especially the insufficient lengths of videos, a lack of diversity in video types and evaluation tasks, and the inappropriateness for evaluating LVU performances. To address the above problems, we propose a new benchmark, called MLVU (Multi-task Long Video Understanding Benchmark), for the comprehensive and in-depth evaluation of LVU. MLVU presents the following critical values: 1) The substantial and flexible extension of video lengths, which enables the benchmark to evaluate LVU performance across a wide range of durations. 2) The inclusion of various video genres, e.g., movies, surveillance footage, egocentric videos, cartoons, game videos, etc., which reflects the models' LVU performances in different scenarios. 3) The development of diversified evaluation tasks, which enables a comprehensive examination of MLLMs' key abilities in long-video understanding. The empirical study with 20 latest MLLMs reveals significant room for improvement in today's technique, as all existing methods struggle with most of the evaluation tasks and exhibit severe performance degradation when handling longer videos. Additionally, it suggests that factors such as context length, image-understanding quality, and the choice of LLM backbone can play critical roles in future advancements. We anticipate that MLVU will advance the research of long video understanding by providing a comprehensive and in-depth analysis of MLLMs.

A key challenge in e-learning environments like Intelligent Tutoring Systems (ITSs) is to induce effective pedagogical policies efficiently. While Deep Reinforcement Learning (DRL) often suffers from sample inefficiency and reward function design difficulty, Apprenticeship Learning(AL) algorithms can overcome them. However, most AL algorithms can not handle heterogeneity as they assume all demonstrations are generated with a homogeneous policy driven by a single reward function. Still, some AL algorithms which consider heterogeneity, often can not generalize to large continuous state space and only work with discrete states. In this paper, we propose an expectation-maximization(EM)-EDM, a general AL framework to induce effective pedagogical policies from given optimal or near-optimal demonstrations, which are assumed to be driven by heterogeneous reward functions. We compare the effectiveness of the policies induced by our proposed EM-EDM against four AL-based baselines and two policies induced by DRL on two different but related tasks that involve pedagogical action prediction. Our overall results showed that, for both tasks, EM-EDM outperforms the four AL baselines across all performance metrics and the two DRL baselines. This suggests that EM-EDM can effectively model complex student pedagogical decision-making processes through the ability to manage a large, continuous state space and adapt to handle diverse and heterogeneous reward functions with very few given demonstrations.

Generalized category discovery presents a challenge in a realistic scenario, which requires the model's generalization ability to recognize unlabeled samples from known and unknown categories. This paper revisits the challenge of generalized category discovery through the lens of information maximization (InfoMax) with a probabilistic parametric classifier. Our findings reveal that ensuring independence between known and unknown classes while concurrently assuming a uniform probability distribution across all classes, yields an enlarged margin among known and unknown classes that promotes the model's performance. To achieve the aforementioned independence, we propose a novel InfoMax-based method, Regularized Parametric InfoMax (RPIM), which adopts pseudo labels to supervise unlabeled samples during InfoMax, while proposing a regularization to ensure the quality of the pseudo labels. Additionally, we introduce novel semantic-bias transformation to refine the features from the pre-trained model instead of direct fine-tuning to rescue the computational costs. Extensive experiments on six benchmark datasets validate the effectiveness of our method. RPIM significantly improves the performance regarding unknown classes, surpassing the state-of-the-art method by an average margin of 3.5%., Comment: Preprint version

Open compound domain adaptation (OCDA) aims to transfer knowledge from a labeled source domain to a mix of unlabeled homogeneous compound target domains while generalizing to open unseen domains. Existing OCDA methods solve the intra-domain gaps by a divide-and-conquer strategy, which divides the problem into several individual and parallel domain adaptation (DA) tasks. Such approaches often contain multiple sub-networks or stages, which may constrain the model's performance. In this work, starting from the general DA theory, we establish the generalization bound for the setting of OCDA. Built upon this, we argue that conventional OCDA approaches may substantially underestimate the inherent variance inside the compound target domains for model generalization. We subsequently present Stochastic Compound Mixing (SCMix), an augmentation strategy with the primary objective of mitigating the divergence between source and mixed target distributions. We provide theoretical analysis to substantiate the superiority of SCMix and prove that the previous methods are sub-groups of our methods. Extensive experiments show that our method attains a lower empirical risk on OCDA semantic segmentation tasks, thus supporting our theories. Combining the transformer architecture, SCMix achieves a notable performance boost compared to the SoTA results.

In controllable generation tasks, flexibly manipulating the generated images to attain a desired appearance or structure based on a single input image cue remains a critical and longstanding challenge. Achieving this requires the effective decoupling of key attributes within the input image data, aiming to get representations accurately. Previous research has predominantly concentrated on disentangling image attributes within feature space. However, the complex distribution present in real-world data often makes the application of such decoupling algorithms to other datasets challenging. Moreover, the granularity of control over feature encoding frequently fails to meet specific task requirements. Upon scrutinizing the characteristics of various generative models, we have observed that the input sensitivity and dynamic evolution properties of the diffusion model can be effectively fused with the explicit decomposition operation in pixel space. This integration enables the image processing operations performed in pixel space for a specific feature distribution of the input image, and can achieve the desired control effect in the generated results. Therefore, we propose FilterPrompt, an approach to enhance the model control effect. It can be universally applied to any diffusion model, allowing users to adjust the representation of specific image features in accordance with task requirements, thereby facilitating more precise and controllable generation outcomes. In particular, our designed experiments demonstrate that the FilterPrompt optimizes feature correlation, mitigates content conflicts during the generation process, and enhances the model's control capability., Comment: Project Page: https://meaoxixi.github.io/FilterPrompt/

In real-world usage, existing GAN image generation tools come up short due to their lack of intuitive interfaces and limited flexibility. To overcome these limitations, we developed CanvasPic, an innovative tool for flexible GAN image generation. Our tool introduces a novel 2D layout design that allows users to intuitively control image attributes based on real-world images. By interacting with the distances between images in the spatial layout, users are able to conveniently control the influence of each attribute on the target image and explore a wide range of generated results. Considering practical application scenarios, a user study involving 24 participants was conducted to compare our tool with existing tools in GAN image generation. The results of the study demonstrate that our tool significantly enhances the user experience, enabling more effective achievement of desired generative results.

Reinforcement Learning from Human Feedback (RLHF) is a widely used framework for the training of language models. However, the process of using RLHF to develop a language model that is well-aligned presents challenges, especially when it comes to optimizing the reward model. Our research has found that existing reward models, when trained using the traditional ranking objective based on human preference data, often struggle to effectively distinguish between responses that are more or less favorable in real-world scenarios. To bridge this gap, our study introduces a novel method to estimate the preference differences without the need for detailed, exhaustive labels from human annotators. Our experimental results provide empirical evidence that incorporating margin values into the training process significantly improves the effectiveness of reward models. This comparative analysis not only demonstrates the superiority of our approach in terms of reward prediction accuracy but also highlights its effectiveness in practical applications.

Diffuse Correlation Spectroscopy (DCS) allows the label-free investigation of microvascular dynamics deep within living tissue. However, common implementations of DCS are currently limited to measurement depths of $\sim 1-1.5cm$, which can limit the accuracy of cerebral hemodynamics measurement. Here we present massively parallelized DCS (pDCS) using novel single photon avalanche detector (SPAD) arrays with up to 500x500 individual channels. The new SPAD array technology can boost the signal-to-noise ratio by a factor of up to 500 compared to single-pixel DCS, or by more than 15-fold compared to the most recent state-of-the-art pDCS demonstrations. Our results demonstrate the first in vivo use of this massively parallelized DCS system to measure cerebral blood flow changes at $\sim 2cm$ depth in human adults. We compared different modes of operation and applied a dual detection strategy, where a secondary SPAD array is used to simultaneously assess the superficial blood flow as a built-in reference measurement. While the blood flow in the superficial scalp tissue showed no significant change during cognitive activation, the deep pDCS measurement showed a statistically significant increase in the derived blood flow index of 8-12% when compared to the control rest state.

Multi-domain generalization (mDG) is universally aimed to minimize the discrepancy between training and testing distributions to enhance marginal-to-label distribution mapping. However, existing mDG literature lacks a general learning objective paradigm and often imposes constraints on static target marginal distributions. In this paper, we propose to leverage a $Y$-mapping to relax the constraint. We rethink the learning objective for mDG and design a new \textbf{general learning objective} to interpret and analyze most existing mDG wisdom. This general objective is bifurcated into two synergistic amis: learning domain-independent conditional features and maximizing a posterior. Explorations also extend to two effective regularization terms that incorporate prior information and suppress invalid causality, alleviating the issues that come with relaxed constraints. We theoretically contribute an upper bound for the domain alignment of domain-independent conditional features, disclosing that many previous mDG endeavors actually \textbf{optimize partially the objective} and thus lead to limited performance. As such, our study distills a general learning objective into four practical components, providing a general, robust, and flexible mechanism to handle complex domain shifts. Extensive empirical results indicate that the proposed objective with $Y$-mapping leads to substantially better mDG performance in various downstream tasks, including regression, segmentation, and classification., Comment: Accepted by CVPR24

Multi-modal large language models(MLLMs) have achieved remarkable progress and demonstrated powerful knowledge comprehension and reasoning abilities. However, the mastery of domain-specific knowledge, which is essential for evaluating the intelligence of MLLMs, continues to be a challenge. Current multi-modal benchmarks for domain-specific knowledge concentrate on multiple-choice questions and are predominantly available in English, which imposes limitations on the comprehensiveness of the evaluation. To this end, we introduce CMMU, a novel benchmark for multi-modal and multi-type question understanding and reasoning in Chinese. CMMU consists of 3,603 questions in 7 subjects, covering knowledge from primary to high school. The questions can be categorized into 3 types: multiple-choice, multiple-response, and fill-in-the-blank, bringing greater challenges to MLLMs. In addition, we propose an evaluation strategy called Positional Error Variance for assessing multiple-choice questions. The strategy aims to perform a quantitative analysis of position bias. We evaluate seven open-source MLLMs along with GPT4-V, Gemini-Pro, and Qwen-VL-Plus. The results demonstrate that CMMU poses a significant challenge to the recent MLLMs. The data and code are available at https://github.com/FlagOpen/CMMU.

Strangeon stars, which are proposed to describe the nature of pulsar-like compact stars, have passed various observational tests. The maximum mass of a non-rotating strangeon star could be high, which implies that the remnants of binary strangeon star mergers could even be long-lived massive strangeon stars. We study rigidly rotating strangeon stars in the slowly rotating approximation, using the Lennard-Jones model for the equation of state. Rotation can significantly increase the maximum mass of strangeon stars with unchanged baryon numbers, enlarging the mass-range of long-lived strangeon stars. During spin-down after merger, the decrease of radius of the remnant will lead to the release of gravitational energy. Taking into account the efficiency of converting the gravitational energy luminosity to the observed X-ray luminosity, we find that the gravitational energy could provide an alternative energy source for the plateau emission of X-ray afterglow. The fitting results of X-ray plateau emission of some short gamma-ray bursts suggest that the magnetic dipole field strength of the remnants can be much smaller than that of expected when the plateau emission is powered only by spin-down luminosity of magnetars., Comment: Accepted by RAA

Considering the appealing distribution gains of distributed antenna systems (DAS) and passive gains of reconfigurable intelligent surface (RIS), a flexible reconfigurable architecture called reconfigurable distributed antenna and reflecting surface (RDARS) is proposed. RDARS encompasses DAS and RIS as two special cases and maintains the advantages of distributed antennas while reducing the hardware cost by replacing some active antennas with low-cost passive reflecting surfaces. In this paper, we present a RDARS-aided uplink multi-user communication system and investigate the system transmission reliability with the newly proposed architecture. Specifically, in addition to the distribution gain and the reflection gain provided by the connection and reflection modes, respectively, we also consider the dynamic mode switching of each element which introduces an additional degree of freedom (DoF) and thus results in a selection gain. As such, we aim to minimize the total sum mean-square-error (MSE) of all data streams by jointly optimizing the receive beamforming matrix, the reflection phase shifts and the channel-aware placement of elements in the connection mode. To tackle this nonconvex problem with intractable binary and cardinality constraints, we propose an inexact block coordinate descent (BCD) based penalty dual decomposition (PDD) algorithm with the guaranteed convergence. Since the PDD algorithm usually suffers from high computational complexity, a low-complexity greedy-search-based alternating optimization (AO) algorithm is developed to yield a semi-closed-form solution with acceptable performance. Numerical results demonstrate the superiority of the proposed architecture compared to the conventional fully passive RIS or DAS. Furthermore, some insights about the practical implementation of RDARS are provided., Comment: 13 pages, 9 figures. This paper has been submitted to IEEE journal for possible publication

We propose an innovative data-driven option pricing methodology that relies exclusively on the dataset of historical underlying asset prices. While the dataset is rooted in the objective world, option prices are commonly expressed as discounted expectations of their terminal payoffs in a risk-neutral world. Bridging this gap motivates us to identify a pricing kernel process, transforming option pricing into evaluating expectations in the objective world. We recover the pricing kernel by solving a utility maximization problem, and evaluate the expectations in terms of a functional optimization problem. Leveraging the deep learning technique, we design data-driven algorithms to solve both optimization problems over the dataset. Numerical experiments are presented to demonstrate the efficiency of our methodology., Comment: 15 pages, 3 figures

Denoising and demosaicking are two fundamental steps in reconstructing a clean full-color video from raw data, while performing video denoising and demosaicking jointly, namely VJDD, could lead to better video restoration performance than performing them separately. In addition to restoration accuracy, another key challenge to VJDD lies in the temporal consistency of consecutive frames. This issue exacerbates when perceptual regularization terms are introduced to enhance video perceptual quality. To address these challenges, we present a new VJDD framework by consistent and accurate latent space propagation, which leverages the estimation of previous frames as prior knowledge to ensure consistent recovery of the current frame. A data temporal consistency (DTC) loss and a relational perception consistency (RPC) loss are accordingly designed. Compared with the commonly used flow-based losses, the proposed losses can circumvent the error accumulation problem caused by inaccurate flow estimation and effectively handle intensity changes in videos, improving much the temporal consistency of output videos while preserving texture details. Extensive experiments demonstrate the leading VJDD performance of our method in term of restoration accuracy, perceptual quality and temporal consistency. Codes and dataset are available at \url{https://github.com/GuoShi28/VJDD}.

To optimize electron energy toward in-situ imaging large bio-samples up to 10-um thickness with nanoscale resolution, we implemented an analytical model based on elastic and inelastic characteristic angles [1]. This model can be used to predict the transverse beam size broadening as a function of electron energy while the probe beam traverses through the sample. As result, the optimal choice of the electron beam energy can be realized. While the sample thickness is less than 10 um, there exists an optimal electron beam energy below 10 MeV regarding a specific sample thickness. However, for samples thicker than 10 um, the optimal beam energy is 10 MeV, and the ultimate resolution could become worse with the increase of the sample thickness.

Recently MLP-based methods have shown strong performance in point cloud analysis. Simple MLP architectures are able to learn geometric features in local point groups yet fail to model long-range dependencies directly. In this paper, we propose Point Deformable Network (PDNet), a concise MLP-based network that can capture long-range relations with strong representation ability. Specifically, we put forward Point Deformable Aggregation Module (PDAM) to improve representation capability in both long-range dependency and adaptive aggregation among points. For each query point, PDAM aggregates information from deformable reference points rather than points in limited local areas. The deformable reference points are generated data-dependent, and we initialize them according to the input point positions. Additional offsets and modulation scalars are learned on the whole point features, which shift the deformable reference points to the regions of interest. We also suggest estimating the normal vector for point clouds and applying Enhanced Normal Embedding (ENE) to the geometric extractors to improve the representation ability of single-point. Extensive experiments and ablation studies on various benchmarks demonstrate the effectiveness and superiority of our PDNet.

In this paper, we investigate a novel reconfigurable distributed antennas and reflecting surface (RDARS) aided multi-user massive MIMO system with imperfect CSI and propose a practical two-timescale (TTS) transceiver design to reduce the communication overhead and computational complexity of the system. In the RDARS-aided system, not only distribution gain but also reflection gain can be obtained by a flexible combination of the distributed antennas and reflecting surface, which differentiates the system from the others and also makes the TTS design challenging. To enable the optimal TTS transceiver design, the achievable rate of the system is first derived in closed-form. Then the TTS design aiming at the weighted sum rate maximization is considered. To solve the challenging non-convex optimization problem with high-order design variables, i.e., the transmit powers and the phase shifts at the RDARS, a block coordinate descent based method is proposed to find the optimal solutions in semi-closed forms iteratively. Specifically, two efficient algorithms are proposed with provable convergence for the optimal phase shift design, i.e., Riemannian Gradient Ascent based algorithm by exploiting the unit-modulus constraints, and Two-Tier Majorization-Minimization based algorithm with closed-form optimal solutions in each iteration. Simulation results validate the effectiveness of the proposed algorithm and demonstrate the superiority of deploying RDARS in massive MIMO systems to provide substantial rate improvement with a significantly reduced total number of active antennas/RF chains and lower transmit power when compared to the DAS and RIS-aided systems., Comment: 13 pages, 6 figures

In this paper, we propose a learning-based image fragment pair-searching and -matching approach to solve the challenging restoration problem. Existing works use rule-based methods to match similar contour shapes or textures, which are always difficult to tune hyperparameters for extensive data and computationally time-consuming. Therefore, we propose a neural network that can effectively utilize neighbor textures with contour shape information to fundamentally improve performance. First, we employ a graph-based network to extract the local contour and texture features of fragments. Then, for the pair-searching task, we adopt a linear transformer-based module to integrate these local features and use contrastive loss to encode the global features of each fragment. For the pair-matching task, we design a weighted fusion module to dynamically fuse extracted local contour and texture features, and formulate a similarity matrix for each pair of fragments to calculate the matching score and infer the adjacent segment of contours. To faithfully evaluate our proposed network, we created a new image fragment dataset through an algorithm we designed that tears complete images into irregular fragments. The experimental results show that our proposed network achieves excellent pair-searching accuracy, reduces matching errors, and significantly reduces computational time. Details, sourcecode, and data are available in our supplementary material., Comment: 14 pages, 16 figures, 4 tables

While recent test-time adaptations exhibit efficacy by adjusting batch normalization to narrow domain disparities, their effectiveness diminishes with realistic mini-batches due to inaccurate target estimation. As previous attempts merely introduce source statistics to mitigate this issue, the fundamental problem of inaccurate target estimation still persists, leaving the intrinsic test-time domain shifts unresolved. This paper delves into the problem of mini-batch degradation. By unraveling batch normalization, we discover that the inexact target statistics largely stem from the substantially reduced class diversity in batch. Drawing upon this insight, we introduce a straightforward tool, Test-time Exponential Moving Average (TEMA), to bridge the class diversity gap between training and testing batches. Importantly, our TEMA adaptively extends the scope of typical methods beyond the current batch to incorporate a diverse set of class information, which in turn boosts an accurate target estimation. Built upon this foundation, we further design a novel layer-wise rectification strategy to consistently promote test-time performance. Our proposed method enjoys a unique advantage as it requires neither training nor tuning parameters, offering a truly hassle-free solution. It significantly enhances model robustness against shifted domains and maintains resilience in diverse real-world scenarios with various batch sizes, achieving state-of-the-art performance on several major benchmarks. Code is available at \url{https://github.com/kiwi12138/RealisticTTA}., Comment: Accepted by AAAI 2024

Data augmentation has been recently leveraged as an effective regularizer in various vision-language deep neural networks. However, in text-to-image synthesis (T2Isyn), current augmentation wisdom still suffers from the semantic mismatch between augmented paired data. Even worse, semantic collapse may occur when generated images are less semantically constrained. In this paper, we develop a novel Semantic-aware Data Augmentation (SADA) framework dedicated to T2Isyn. In particular, we propose to augment texts in the semantic space via an Implicit Textual Semantic Preserving Augmentation ($ITA$), in conjunction with a specifically designed Image Semantic Regularization Loss ($L_r$) as Generated Image Semantic Conservation, to cope well with semantic mismatch and collapse. As one major contribution, we theoretically show that $ITA$ can certify better text-image consistency while $L_r$ regularizing the semantics of generated images would avoid semantic collapse and enhance image quality. Extensive experiments validate that SADA enhances text-image consistency and improves image quality significantly in T2Isyn models across various backbones. Especially, incorporating SADA during the tuning process of Stable Diffusion models also yields performance improvements., Comment: Accepted by AAAI24

Objective To solve major clinical natural language processing (NLP) tasks using a unified text-to-text learning architecture based on a generative large language model (LLM) via prompt tuning. Methods We formulated 7 key clinical NLP tasks as text-to-text learning and solved them using one unified generative clinical LLM, GatorTronGPT, developed using GPT-3 architecture and trained with up to 20 billion parameters. We adopted soft prompts (i.e., trainable vectors) with frozen LLM, where the LLM parameters were not updated (i.e., frozen) and only the vectors of soft prompts were updated, known as prompt tuning. We added additional soft prompts as a prefix to the input layer, which were optimized during the prompt tuning. We evaluated the proposed method using 7 clinical NLP tasks and compared them with previous task-specific solutions based on Transformer models. Results and Conclusion The proposed approach achieved state-of-the-art performance for 5 out of 7 major clinical NLP tasks using one unified generative LLM. Our approach outperformed previous task-specific transformer models by ~3% for concept extraction and 7% for relation extraction applied to social determinants of health, 3.4% for clinical concept normalization, 3.4~10% for clinical abbreviation disambiguation, and 5.5~9% for natural language inference. Our approach also outperformed a previously developed prompt-based machine reading comprehension (MRC) model, GatorTron-MRC, for clinical concept and relation extraction. The proposed approach can deliver the ``one model for all`` promise from training to deployment using a unified generative LLM.

Accurate channel state information (CSI) is essential for downlink precoding in frequency division duplexing (FDD) massive multiple-input multiple-output (MIMO) systems with orthogonal frequency-division multiplexing (OFDM). However, obtaining CSI through feedback from the user equipment (UE) becomes challenging with the increasing scale of antennas and subcarriers and leads to extremely high CSI feedback overhead. Deep learning-based methods have emerged for compressing CSI but these methods generally require substantial collected samples and thus pose practical challenges. Moreover, existing deep learning methods also suffer from dramatically growing feedback overhead owing to their focus on full-dimensional CSI feedback. To address these issues, we propose a low-overhead Incorporation-Extrapolation based Few-Shot CSI feedback Framework (IEFSF) for massive MIMO systems. An incorporation-extrapolation scheme for eigenvector-based CSI feedback is proposed to reduce the feedback overhead. Then, to alleviate the necessity of extensive collected samples and enable few-shot CSI feedback, we further propose a knowledge-driven data augmentation (KDDA) method and an artificial intelligence-generated content (AIGC) -based data augmentation method by exploiting the domain knowledge of wireless channels and by exploiting a novel generative model, respectively. Experimental results based on the DeepMIMO dataset demonstrate that the proposed IEFSF significantly reduces CSI feedback overhead by 64 times compared with existing methods while maintaining higher feedback accuracy using only several hundred collected samples., Comment: 16 pages, 12 figures, 5 tables. Accepted by IEEE Transactions on Wireless Communications

Nowadays, it is common to release audio content to the public. However, with the rise of voice cloning technology, attackers have the potential to easily impersonate a specific person by utilizing his publicly released audio without any permission. Therefore, it becomes significant to detect any potential misuse of the released audio content and protect its timbre from being impersonated. To this end, we introduce a novel concept, "Timbre Watermarking", which embeds watermark information into the target individual's speech, eventually defeating the voice cloning attacks. To ensure the watermark is robust to the voice cloning model's learning process, we design an end-to-end voice cloning-resistant detection framework. The core idea of our solution is to embed and extract the watermark in the frequency domain in a temporally invariant manner. To acquire generalization across different voice cloning attacks, we modulate their shared process and integrate it into our framework as a distortion layer. Experiments demonstrate that the proposed timbre watermarking can defend against different voice cloning attacks, exhibit strong resistance against various adaptive attacks (e.g., reconstruction-based removal attacks, watermark overwriting attacks), and achieve practicality in real-world services such as PaddleSpeech, Voice-Cloning-App, and so-vits-svc. In addition, ablation studies are also conducted to verify the effectiveness of our design. Some audio samples are available at https://timbrewatermarking.github.io/samples., Comment: NDSS 2024

Point cloud registration (PCR) is an essential task in 3D vision. Existing methods achieve increasingly higher accuracy. However, a large proportion of non-overlapping points in point cloud registration consume a lot of computational resources while negatively affecting registration accuracy. To overcome this challenge, we introduce a novel Efficient Point Cloud Registration via Dynamic Iteration framework, DIPR, that makes the neural network interactively focus on overlapping points based on sparser input points. We design global and local registration stages to achieve efficient course-tofine processing. Beyond basic matching modules, we propose the Refined Nodes to narrow down the scope of overlapping points by using adopted density-based clustering to significantly reduce the computation amount. And our SC Classifier serves as an early-exit mechanism to terminate the registration process in time according to matching accuracy. Extensive experiments on multiple datasets show that our proposed approach achieves superior registration accuracy while significantly reducing computational time and GPU memory consumption compared to state-of-the-art methods.